Pith. sign in

REVIEW 3 major objections 6 minor 45 references

Pairing a configuration-driven FDTR code package with procedural agent skills lets language-model agents run reliable thermal analyses from plain-language requests.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-31 14:51 UTC pith:CLAQGDTZ

load-bearing objection Solid agent-systems paper for FDTR: the ablations land, the reliability claim is real on their benchmark, and the main caveat is external validity—not internal collapse. the 3 major comments →

arxiv 2607.28200 v1 pith:CLAQGDTZ submitted 2026-07-30 physics.app-ph cond-mat.mtrl-scics.AI

Vibe-FDTR: An agent-oriented framework for reproducible frequency-domain thermoreflectance data analysis

Fuwei Yang , Weiheng Li , Bai Song This is my paper
classification physics.app-ph cond-mat.mtrl-scics.AI
keywords thermal measurementfrequency-domain thermoreflectancedata analysisLLM agentagent-oriented frameworksensitivity analysisuncertainty propagation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Frequency-domain thermoreflectance (FDTR) can measure micro- and nanoscale thermal properties, but turning raw pump-probe signals into trustworthy numbers is a multi-step inverse problem that usually needs expert judgment and is easy to get subtly wrong. This paper introduces Vibe-FDTR, a framework that couples a domain FDTR software package—which enforces physical and parametric consistency through configuration files—with procedural “skills” that tell an LLM agent how to turn a natural-language request into ordered, checkable analysis steps. On a controlled two-level benchmark (synthetic single-step tasks and multi-step real-data workflows on gold-coated graphite), agents using the full framework succeed on essentially every run, while stripping the skills or the domain package collapses reliability, especially on complex workflows, and also burns far more time and API cost. An optional expert mode can plan sensitivity and uncertainty studies and recommend what to fit when the request is underspecified. The authors’ claim is that packaging validated domain code and expert procedure into agent skills is a practical path to lower-barrier, reproducible thermal metrology.

Core claim

The paper claims that LLM agents can perform reliable, reproducible FDTR data analysis from natural language when given both a configuration-driven FDTR code package that enforces physical consistency and procedural agent skills that map user intent to verifiable steps. On their L1/L2 benchmark this combination yields 100% and 98.9% success rates, versus 91.4%/36.7% with skills removed and 38.6%/0% with the domain package also removed, while cutting cost by about 88% and runtime by more than 60% relative to the code-only agent.

What carries the argument

Vibe-FDTR’s two-layer core: a configuration-driven FDTR code package (shared parameters, validation, pipelines) plus progressive-disclosure procedural agent skills that route tasks, build configs, and run fitting, sensitivity, uncertainty, and iterative workflows.

Load-bearing premise

Success is defined as matching the authors’ synthetic ground truth or human-expert fits on the same FDTR backend within preset tolerances under one fixed model and harness, which is taken to stand for trustworthy real-world analysis.

What would settle it

Rerun the same L1/L2 task suite with independent human analysts using a different FDTR backend, or with another frontier LLM harness, and check whether full Vibe-FDTR still near-perfectly recovers the external reference values while the ablated setups remain far worse.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Well-specified FDTR post-processing can be requested in natural language with near-perfect repeatability on the authors’ task class.
  • Multi-step real-data workflows (batch temperature fits, iterative spot-size and parameter transfer) become much more reliable once procedural skills guide the domain package.
  • API cost and wall time for agent-driven FDTR analysis drop sharply versus letting the model explore raw code.
  • The same encapsulation pattern can be extended to related techniques such as TDTR and, later, instrument control.
  • Optional expert mode can turn underspecified planning questions into sensitivity/uncertainty-backed fitting recommendations, though not yet as a full substitute for human experts.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Other laser pump-probe and electrothermal methods with similar multilayer inverse fits are natural next targets for the same code-plus-skills pattern.
  • Because L2 truth is same-backend expert output, cross-lab or closed-source-model transfer remains an open test of whether the reliability claim generalizes.
  • Failure modes in the ablations (CLI misuse, parameter-name mixups, stalled model derivation) suggest that progressive skill disclosure is doing most of the work once the physics package exists.
  • Closing the loop from analysis skills to live acquisition would be the concrete step from “copilot for fitting” to autonomous metrology as the conclusion sketches.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The manuscript introduces Vibe-FDTR, an agent-oriented framework that couples a configuration-driven FDTR analysis package (enforcing physical/parametric consistency) with procedural LLM agent skills that map natural-language requests to verifiable analysis steps, plus an optional expert mode for underspecified planning. Reliability is quantified on a two-level benchmark: seven synthetic single-step L1 tasks and nine multi-step L2 workflows on real Au/graphite FDTR data, each run 10 times in a containerized OpenCode/DeepSeek-V4-Pro harness with explicit success criteria (timeout, required outputs, numerical tolerances). Vibe-FDTR reports 100% (L1) and 98.9% (L2) success, versus 91.4%/36.7% with skills ablated (Code-agent) and 38.6%/0% with the domain package also omitted (Agent-only), together with ~87.7% lower API cost and >60% shorter wall time versus Code-agent. Failure modes, cost/time breakdowns, and qualitative expert-mode traces (Table 2, Fig. 8) are documented.

Significance. If the reported rates hold under the stated protocol, this is a concrete and useful systems contribution for thermal metrology: it shows that encapsulating validated FDTR code and expert workflow into agent skills can make multi-step nonlinear fitting, sensitivity, and uncertainty analysis reliably executable from natural language, with large efficiency gains over code-only agent use. Strengths include a clear ablation design isolating package versus skills (§4.2, Fig. 6–7), repeated-run statistics (n=10), concrete failure-mode analysis (§5.2), and honest self-limitation of expert mode (§5.4). The work is timely given rising agent tooling in experimental science and could lower the barrier for FDTR/TDTR practitioners if the software is released as promised.

major comments (3)
  1. [§4.2, §5.1, Abstract, Conclusion] §4.2 and §5.1–5.2: L2 “success” is defined as numerical agreement with human-expert fits on the same FDTR backend (and L1 with the synthetic forward-model parameters) within preset tolerances under a single harness/model. That is a valid engineering reliability metric, but the abstract and conclusion frame the result as enabling “trustworthy thermal metrology” and a route to “fully autonomous” analysis. Please scope the central claim explicitly to controlled-benchmark agent reliability (reproducible execution matching a validated backend), and separate it from independent physical validation, cross-lab reproducibility, or model-agnostic correctness. Without that distinction, the strongest wording overreaches what Table 1 and the evaluation protocol establish.
  2. [§4.2, §5.1, Abstract] §4.2 and §5.1: All quantitative rates use DeepSeek-V4-Pro via OpenCode in the authors’ Docker setup. The L1→L2 collapse of Code-agent/Agent-only and the cost/time gains are therefore substrate-conditioned. At minimum, state this limitation prominently near the headline numbers and discuss how sensitive the skill layer is expected to be to other frontier models; ideally add a small multi-model spot check on a subset of L2 tasks. Otherwise the generalization implied by “LLM agents” in the abstract is not supported.
  3. [Data and code availability; §4.2] Data and code availability: the manuscript states that source, skills, benchmark runner, and traces “will be” deposited/available, but does not provide frozen commit hashes, configuration schemas, or the numerical tolerance tables used for success scoring. For a methods paper whose load-bearing claim is reproducibility of agent runs, release (or staged anonymous release) of the package, skill documents, task prompts, and scoring criteria is essential so that the 100%/98.9% figures can be audited independently.
minor comments (6)
  1. [Figure 6] Fig. 6 encodes success counts only as a color scale without a numeric legend per cell; adding the integer counts (or a supplementary table) would make the 98.9% / 36.7% / 0% aggregates easier to verify task-by-task.
  2. [Table 1, §4.2] Table 1 lists main targets compactly but does not state the numerical tolerances or which signal channels/windows define success for each ID; a short supplementary table would strengthen the evaluation protocol.
  3. [§2.3] Eqs. (9)–(11): clarify whether amplitude, phase, or complex residual enters R and Var(Y), and whether weights differ across f-sweep versus beam-offset fits; this affects interpretation of L1-U01 and L2 uncertainty tasks.
  4. [§5.4, Table 2] Expert-mode E tasks (Table 2) are valuable but purely qualitative; a brief rubric (e.g., assumption completeness, sensitivity coverage, risk disclosure) would make §5.4 less anecdotal without claiming quantitative success rates.
  5. [§1, Figure 2] Minor prose/typo issues: “andharnesses,” “Very recently… andharnesses,” spacing in “frequency-domainthermoreflectance,” and inconsistent κ_i/κ_o versus κ_r/κ_z notation between text and Fig. 2.
  6. [§1] Related-work placement: neural-network TDTR/FDTR inverse papers [38–40] are cited; a sentence contrasting agentic workflow orchestration with pure inverse surrogates would sharpen novelty for non-AI readers.

Circularity Check

0 steps flagged

No significant circularity: benchmark success rates are empirical agent-vs-reference outcomes, not quantities forced by definition or self-citation.

full rationale

Vibe-FDTR is a systems/methods paper whose load-bearing claims are measured success rates, cost, and runtime of LLM agents on a controlled L1/L2 benchmark under ablations (skills off; package off). L1 ground truth is the synthetic forward-model parameters used to generate the data; L2 ground truth is human-expert fits on the same FDTR backend within preset numerical tolerances (§4.2, Table 1). Matching those references is an independent execution test of the agent workflow, not a fitted quantity renamed as a prediction, nor a result defined in terms of itself. The FDTR thermal model (Eqs. 1–11), sensitivity, and uncertainty formulas are standard multilayer heat-diffusion machinery cited to the literature, not derived circularly from the agent results. Self-citations to the authors’ prior FDTR experiments supply sample context and setup, not a uniqueness theorem that forces the success-rate claim. Expert-mode outputs are qualitative recommendations with acknowledged limits (§5.4), not circular first-principles predictions. External-validity caveats (same backend, one harness/model) affect generalization, not internal circularity. No step reduces a claimed prediction to its inputs by construction.

Axiom & Free-Parameter Ledger

3 free parameters · 5 axioms · 2 invented entities

The central claim is empirical systems performance, not a first-principles derivation. It rests on standard multilayer FDTR heat-diffusion modeling, a particular success definition tied to the authors’ backend and human references, a chosen LLM/harness, and the premise that skill documents plus config validation encode expert procedure sufficiently for the benchmark distribution.

free parameters (3)
  • L1/L2 numerical success tolerances and time limits (600 s / 900 s) = Not numerically listed beyond time limits 600 s (L1) and 900 s (L2)
    Binary success depends on author-chosen agreement windows and cutoffs (§4.2); shifting tolerances would change reported rates.
  • Expert-mode default assumptions (e.g., spot size ~3 µm, fixed G values, Si-like film properties in E-7) = Case-dependent (e.g., G1 = 200 MW m−2 K−1 in E-7 trace)
    Underspecified planning tasks inject hand-chosen nominals that shape sensitivity/uncertainty conclusions and recommendations (Table 2, Fig. 8).
  • Human-expert L2 reference fits on same backend
    L2 ground truth is not an external metrology standard; it is expert analysis with the same code path, so the metric partly measures agent–expert agreement under shared model assumptions.
axioms (5)
  • domain assumption Multilayer Fourier heat diffusion with transversely isotropic layers, Gaussian pump/probe averaging, and transfer-matrix interfaces adequately models the FDTR signals used for fitting.
    Section 2.2 presents Eqs. (1)–(7) as the fitting model; package and benchmark assume this physics class.
  • domain assumption Normalized sensitivity S_x = ∂ln y/∂ln x and the local Jacobian error-propagation formula (Eq. 11) are sufficient reliability diagnostics for parameter identifiability.
    Section 2.3; paper notes locality and missing global-optima errors, yet skills and expert mode rely on these tools.
  • ad hoc to paper Matching synthetic forward parameters (L1) or same-backend human expert results within tolerances constitutes a successful, trustworthy analysis run.
    Evaluation protocol §4.2 defines the headline success rates; this is a benchmark convention, not a universal metrology axiom.
  • ad hoc to paper Procedural skill documents plus configuration validation can encode FDTR expert workflow well enough that an LLM need not reverse-engineer source code.
    Core design claim of §3; supported by ablations but still an architectural premise about skill completeness and LLM compliance.
  • ad hoc to paper DeepSeek-V4-Pro via OpenCode in the authors’ Docker setup is a representative agent substrate for the reported rates.
    §4.2; paper itself notes expert-mode quality depends strongly on the underlying LLM (§5.4).
invented entities (2)
  • Vibe-FDTR framework (code layer + procedural guidance skills + optional expert skill) no independent evidence
    purpose: Translate natural-language FDTR intents into validated, multi-step analyses and optional experimental planning.
    Primary contribution; existence is demonstrated by description and benchmark behavior, but independent public artifact evidence is not yet linked in the preprint.
  • Two-level Vibe-FDTR benchmark (L1 synthetic, L2 Au/graphite multi-step, plus E underspecified tasks) no independent evidence
    purpose: Quantify agent reliability, cost, and time under controlled prompts and ablations.
    Custom evaluation suite; necessary for the claim but author-defined rather than a community standard.

pith-pipeline@v1.2.0-daily-grok45 · 22020 in / 3679 out tokens · 65683 ms · 2026-07-31T14:51:05.515774+00:00 · methodology

0 comments
read the original abstract

Frequency-domain thermoreflectance (FDTR) is a laser pump-probe technique widely used to measure thermal properties at the micro- and nanoscale; however, it relies on a complex data analysis procedure that demands substantial domain expertise and is susceptible to subtle human errors. Here, we present Vibe-FDTR, an agent-oriented framework that enables large language model (LLM) agents to perform reliable and reproducible FDTR analyses directly from natural language requests. This framework couples a configuration-driven FDTR code package, which enforces physical and parametric consistency, with procedural agent skills that translate user intentions into organized and verifiable analysis steps. We evaluate Vibe-FDTR using a controlled benchmark with two levels: synthetic single-step tasks and real-data multi-step tasks based on measurements of gold-coated graphite samples. Across the two levels, agents using Vibe-FDTR achieve success rates of 100% and 98.9%, respectively. In sharp contrast, ablating skills (Code-agent) reduces performance to 91.4% and 36.7%, which drops further to 38.6% and 0% when the domain package is also omitted (Agent-only). Beyond success rate, Vibe-FDTR also reduces computational cost by 87.7% relative to the Code-agent variant and cuts execution time by more than 60%. Finally, an optional expert mode supports experimental planning via autonomous sensitivity and uncertainty evaluations, and formulates physically grounded recommendations for underspecified tasks. These results demonstrate that encapsulating domain code and expert knowledge into agent skills offers a promising route toward low-barrier, autonomous, and trustworthy thermal metrology.

Figures

Figures reproduced from arXiv: 2607.28200 by Bai Song, Fuwei Yang, Weiheng Li.

Figure 8
Figure 8. Figure 8: First, the agent frames the task as a design-only problem and addresses the miss [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 1
Figure 1. Figure 1: Basic principles of the FDTR technique. (a) Schematic of FDTR experimental setup. (b) Representative beam-offset phase signals at f = 1 MHz measured on an Au/graphite sample at three different temperatures. (c) Representative f-sweep signals measured on the same sample. The insets show the schematics of the two measurement modes. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Schematic showing the concept of Vibe-FDTR in comparison with traditional manual analysis process. 19 [PITH_FULL_IMAGE:figures/full_fig_p019_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Core architecture of Vibe-FDTR which is composed of a procedural guidance layer and a code layer. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Optional expert mode workflow of Vibe-FDTR. 21 [PITH_FULL_IMAGE:figures/full_fig_p021_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Design of the evaluation environment for benchmark tasks. 22 [PITH_FULL_IMAGE:figures/full_fig_p022_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Number of successful runs of L1 and L2 benchmark tasks for Vibe-FDTR and the two reference variants. Each task is repeatedly run for 10 times. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Average time and cost per run for L1 and L2 benchmark tasks using Vibe-FDTR and the two reference variants. 24 [PITH_FULL_IMAGE:figures/full_fig_p024_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Representative execution trace under the optional expert mode of Vibe-FDTR. The E-7 task is demonstrated as an example, and the agent outputs are selected and reorganized for clearance. 25 [PITH_FULL_IMAGE:figures/full_fig_p025_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

45 extracted references · 1 canonical work pages

  1. [1]

    Nanoscale Thermal Transport

    D. G. Cahill, W. K. Ford, K. E. Goodson, et al. “Nanoscale Thermal Transport”. In:Journal of Applied Physics93.2 (Jan. 2003), pp. 793–818

    2003

  2. [2]

    Emerging Challenges and Materials for Thermal Man- agement of Electronics

    Arden L. Moore and Li Shi. “Emerging Challenges and Materials for Thermal Man- agement of Electronics”. In:Materials Today17.4 (May 2014), pp. 163–174. (Visited on 01/16/2022)

    2014

  3. [3]

    Near-field ra- diative thermal transport: From theory to experiment

    Bai Song, Anthony Fiorino, Edgar Meyhofer, and Pramod Reddy. “Near-field ra- diative thermal transport: From theory to experiment”. In:AIP Advances5.5 (Apr. 2015), p. 053503.doi:10.1063/1.4919048.url:https://doi.org/10.1063/1. 4919048(visited on 07/28/2026)

    work page doi:10.1063/1.4919048.url:https://doi.org/10.1063/1 2015

  4. [4]

    Measurement Techniques for Thermal Conductivity and Interfacial Thermal Con- ductance of Bulk and Thin Film Materials

    Dongliang Zhao, Xin Qian, Xiaokun Gu, Saad Ayub Jajja, and Ronggui Yang. “Measurement Techniques for Thermal Conductivity and Interfacial Thermal Con- ductance of Bulk and Thin Film Materials”. In:Journal of Electronic Packaging 138.040802 (Oct. 2016). (Visited on 06/25/2026)

    2016

  5. [5]

    Interfacial Thermal Resis- tance: Past, Present, and Future

    Jie Chen, Xiangfan Xu, Jun Zhou, and Baowen Li. “Interfacial Thermal Resis- tance: Past, Present, and Future”. In:Reviews of Modern Physics94.2 (Apr. 2022), p. 025002. (Visited on 04/25/2022)

    2022

  6. [6]

    Thermal Conductivity Measurement from 30 to 750 K: The 3ω Method

    David G. Cahill. “Thermal Conductivity Measurement from 30 to 750 K: The 3ω Method”. In:Review of Scientific Instruments61.2 (Feb. 1990), pp. 802–808. (Vis- ited on 06/10/2022)

    1990

  7. [7]

    MEASURINGTHETHERMALCONDUCTIVITYOFTHINFILMS: 3 OMEGA AND RELATED ELECTROTHERMAL METHODS

    ChrisDames.“MEASURINGTHETHERMALCONDUCTIVITYOFTHINFILMS: 3 OMEGA AND RELATED ELECTROTHERMAL METHODS”. In:Annual Re- view of Heat Transfer16.1 (2013), pp. 7–49. (Visited on 06/25/2026)

    2013

  8. [8]

    Scanning Thermal Microscopy

    A. Majumdar. “Scanning Thermal Microscopy”. In:Annual Review of Materials Science29.1 (Aug. 1999), pp. 505–585. (Visited on 03/08/2022)

    1999

  9. [9]

    Thermometry and Thermal Transport in Micro/Nanoscale Solid-State Devices and Structures

    David G. Cahill, Kenneth Goodson, and Arunava Majumdar. “Thermometry and Thermal Transport in Micro/Nanoscale Solid-State Devices and Structures”. In: Journal of Heat Transfer124.2 (Dec. 2001), pp. 223–241. (Visited on 06/25/2026)

    2001

  10. [10]

    Thermal Transport Measurements of Nanostruc- tures Using Suspended Micro-Devices

    Sunmi Shin and Renkun Chen. “Thermal Transport Measurements of Nanostruc- tures Using Suspended Micro-Devices”. In:Nanoscale Energy Transport: Emerging Phenomena, Methods and Applications. IOP Publishing, Mar. 2020. (Visited on 06/25/2026)

    2020

  11. [11]

    Big MEMS for Thermal Mea- surement

    Haiyu He, Yuxi Wang, Zhiyao Jiang, and Bai Song. “Big MEMS for Thermal Mea- surement”. In:ASME Journal of Heat and Mass Transfer(Sept. 2024), pp. 1–28. (Visited on 11/17/2024). 26

    2024

  12. [12]

    Big-MEMS for high thermal conductivity measurement

    Haiyu He, Yuxi Wang, and Bai Song. “Big-MEMS for high thermal conductivity measurement”. en. In:Review of Scientific Instruments96.2 (Feb. 2025).doi:10. 1063/5.0245017.url:https://pubs.aip.org/aip/rsi/article/96/2/024902/ 3333957/Big-MEMS-for-high-thermal-conductivity-measurement(visited on 09/15/2025)

    2025

  13. [13]

    Raman Spectroscopy for Thermal Transport Characterization: Prin- ciples, Techniques, and Applications

    Michel Kazan. “Raman Spectroscopy for Thermal Transport Characterization: Prin- ciples, Techniques, and Applications”. In:Journal of Applied Physics139.8 (Feb. 2026), p. 081101. (Visited on 06/25/2026)

    2026

  14. [14]

    Raman- Based Nanoscale Thermal Transport Characterization: A Critical Review

    Shen Xu, Aoran Fan, Haidong Wang, Xing Zhang, and Xinwei Wang. “Raman- Based Nanoscale Thermal Transport Characterization: A Critical Review”. In:In- ternational Journal of Heat and Mass Transfer154 (June 2020), p. 119751. (Visited on 06/25/2026)

    2020

  15. [15]

    Tutorial: Time-domain Thermore- flectance (TDTR) for Thermal Property Characterization of Bulk and Thin Film Materials

    Puqing Jiang, Xin Qian, and Ronggui Yang. “Tutorial: Time-domain Thermore- flectance (TDTR) for Thermal Property Characterization of Bulk and Thin Film Materials”. In:Journal of Applied Physics124.16 (Oct. 2018), p. 161103

    2018

  16. [16]

    An Instrumentation Guide to Measuring Thermal Conductiv- ity Using Frequency Domain Thermoreflectance (FDTR)

    Dylan J. Kirsch, Joshua Martin, Ronald Warzoha, Mark McLean, Donald Windover, and Ichiro Takeuchi. “An Instrumentation Guide to Measuring Thermal Conductiv- ity Using Frequency Domain Thermoreflectance (FDTR)”. In:Review of Scientific Instruments95.10 (Oct. 2024), p. 103006. (Visited on 11/04/2024)

    2024

  17. [17]

    Nanoscale Thermal Transport. II. 2003-2012

    David G. Cahill, Paul V. Braun, Gang Chen, et al. “Nanoscale Thermal Transport. II. 2003-2012”. In:Applied Physics Reviews1.1 (Mar. 2014)

    2003

  18. [18]

    Spatially Resolved Thermoreflectance Techniques for Thermal Conductivity Measurements from the Nanoscale to the Mesoscale

    David H. Olson, Jeffrey L. Braun, and Patrick E. Hopkins. “Spatially Resolved Thermoreflectance Techniques for Thermal Conductivity Measurements from the Nanoscale to the Mesoscale”. In:Journal of Applied Physics126.15 (2019)

    2019

  19. [19]

    Ultrahigh Thermal Con- ductivity in Isotope-Enriched Cubic Boron Nitride

    Ke Chen, Bai Song, Navaneetha K. Ravichandran, et al. “Ultrahigh Thermal Con- ductivity in Isotope-Enriched Cubic Boron Nitride”. In:Science367.6477 (Jan. 2020), pp. 555–559

    2020

  20. [20]

    Metallicθ-Phase Tantalum Nitride Has a Thermal Conductivity Triple That of Copper

    Suixuan Li, Chuanjin Su, Zihao Qin, et al. “Metallicθ-Phase Tantalum Nitride Has a Thermal Conductivity Triple That of Copper”. In:Science391.6786 (Feb. 2026), pp. 707–711. (Visited on 06/25/2026)

    2026

  21. [21]

    A Frequency-Domain Ther- moreflectance Method for the Characterization of Thermal Properties

    Aaron J. Schmidt, Ramez Cheaito, and Matteo Chiesa. “A Frequency-Domain Ther- moreflectance Method for the Characterization of Thermal Properties”. In:Review of Scientific Instruments80.9 (Sept. 2009), p. 094901. (Visited on 06/08/2022)

    2009

  22. [22]

    Unusual High Thermal Conductivity in Boron Arsenide Bulk Crystals

    Fei Tian, Bai Song, Xi Chen, et al. “Unusual High Thermal Conductivity in Boron Arsenide Bulk Crystals”. In:Science361.6402 (Aug. 2018), pp. 582–585. (Visited on 11/27/2022). 27

    2018

  23. [23]

    Thermal Transport in a 2D Amorphous Material

    Yuxi Wang, Nianjie Liang, Xingxing Zhang, et al. “Thermal Transport in a 2D Amorphous Material”. In:Physical Review X15.3 (Sept. 2025), p. 031077. (Visited on 01/12/2026)

    2025

  24. [24]

    Thermal Transport in Rhombohedral Boron Nitride

    Fuwei Yang, Wenjiang Zhou, Tian Gu, et al. “Thermal Transport in Rhombohedral Boron Nitride”. In:Physical Review B112.11 (Sept. 2025), p. 115422. (Visited on 09/21/2025)

    2025

  25. [25]

    Observation of Solid-State Bidirectional Thermal Conductivity Switching in Antiferroelectric Lead Zirconate (PbZrO3)

    Kiumars Aryana, John A. Tomko, Ran Gao, et al. “Observation of Solid-State Bidirectional Thermal Conductivity Switching in Antiferroelectric Lead Zirconate (PbZrO3)”. In:Nature Communications13.1 (Mar. 2022), p. 1573. (Visited on 06/25/2026)

    2022

  26. [26]

    Electric Field–Dependent Phonon Spectrum and Heat Conduction in Ferroelectrics

    Brandi L. Wooten, Ryo Iguchi, Ping Tang, et al. “Electric Field–Dependent Phonon Spectrum and Heat Conduction in Ferroelectrics”. In:Science Advances9.5 (Feb. 2023), eadd7194. (Visited on 06/25/2026)

    2023

  27. [27]

    Measurements of Thermal Resistance Across Buried Interfaces with Frequency-Domain Thermore- flectance and Microscale Confinement

    Ronald J. Warzoha, Adam A. Wilson, Brian F. Donovan, et al. “Measurements of Thermal Resistance Across Buried Interfaces with Frequency-Domain Thermore- flectance and Microscale Confinement”. In:ACS Applied Materials & Interfaces 16.31 (Aug. 2024), pp. 41633–41641. (Visited on 10/03/2024)

    2024

  28. [28]

    Ultrahigh Thermal Conduc- tance across Superlubric Interfaces in Twisted Graphite

    Fuwei Yang, Wenjiang Zhou, Zhibin Zhang, et al. “Ultrahigh Thermal Conduc- tance across Superlubric Interfaces in Twisted Graphite”. In:Physical Review Letters 134.14 (Apr. 2025), p. 146302. (Visited on 04/14/2025)

    2025

  29. [29]

    Anomalous Enhancement of Ther- mal Conduction across Twisted van Der Waals Heterointerfaces

    Yufeng Zhang, Yanzheng Du, Xiao Wan, et al. “Anomalous Enhancement of Ther- mal Conduction across Twisted van Der Waals Heterointerfaces”. In:Proceedings of the National Academy of Sciences123.9 (Mar. 2026), e2531049123. (Visited on 03/02/2026)

    2026

  30. [30]

    Analysis of Heat Flow in Layered Structures for Time-Domain Thermoreflectance

    David G. Cahill. “Analysis of Heat Flow in Layered Structures for Time-Domain Thermoreflectance”. In:Review of Scientific Instruments75.12 (2004), pp. 5119– 5122

    2004

  31. [31]

    Time-Domain Thermoreflectance (TDTR) Mea- surements of Anisotropic Thermal Conductivity Using a Variable Spot Size Ap- proach

    P. Jiang, X. Qian, and R. Yang. “Time-Domain Thermoreflectance (TDTR) Mea- surements of Anisotropic Thermal Conductivity Using a Variable Spot Size Ap- proach”. In:Review of Scientific Instruments88.7 (July 2017), p. 074901

    2017

  32. [32]

    A New Spatial-DomainThermoreflectanceMethodtoMeasureaBroadRangeofAnisotropic in-Plane Thermal Conductivity

    Puqing Jiang, Dihui Wang, Zeyu Xiang, Ronggui Yang, and Heng Ban. “A New Spatial-DomainThermoreflectanceMethodtoMeasureaBroadRangeofAnisotropic in-Plane Thermal Conductivity”. In:International Journal of Heat and Mass Trans- fer191 (Aug. 2022), p. 122849. (Visited on 10/24/2022)

    2022

  33. [33]

    DeepSeek-AI, Anyi Xu, Bangcai Lin, et al.DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence. 2026. arXiv:2606.19348 [cs.CL].url:https: //arxiv.org/abs/2606.19348. 28

    arXiv 2026

  34. [34]

    Towards End-to-End Automa- tion of AI Research

    Chris Lu, Cong Lu, Robert Tjarko Lange, et al. “Towards End-to-End Automa- tion of AI Research”. In:Nature651.8107 (Mar. 2026), pp. 914–919. (Visited on 03/30/2026)

    2026

  35. [35]

    Science Acceleration and Accessibility with Self-Driving Labs

    Richard B. Canty, Jeffrey A. Bennett, Keith A. Brown, et al. “Science Acceleration and Accessibility with Self-Driving Labs”. In:Nature Communications16.1 (Apr. 2025), p. 3856. (Visited on 06/25/2026)

    2025

  36. [36]

    Evaluating Large Language Model Agents for Automation of Atomic Force Microscopy

    Indrajeet Mandal, Jitendra Soni, Mohd Zaki, et al. “Evaluating Large Language Model Agents for Automation of Atomic Force Microscopy”. In:Nature Communi- cations16.1 (Oct. 2025), p. 9104. (Visited on 05/20/2026)

    2025

  37. [37]

    J 3SPM AI: An Integrated Open-Source Platform for AI-assisted Image Analysis and Image-Guided Workflows in Scanning Probe Microscopy

    SangHeon Lee. “J 3SPM AI: An Integrated Open-Source Platform for AI-assisted Image Analysis and Image-Guided Workflows in Scanning Probe Microscopy”. In: Micron204 (May 2026), p. 104017. (Visited on 06/25/2026)

    2026

  38. [38]

    Machine Learning-Based Data Pro- cessing Technique for Time-Domain Thermoreflectance (TDTR) Measurements

    Yu Pang, Puqing Jiang, and Ronggui Yang. “Machine Learning-Based Data Pro- cessing Technique for Time-Domain Thermoreflectance (TDTR) Measurements”. In: Journal of Applied Physics130.8 (Aug. 2021), p. 084901. (Visited on 05/09/2025)

    2021

  39. [39]

    Estimating Depth-Directional Thermal Conductivity Profiles Using Neural Network with Dropout in Frequency-Domain Thermoreflectance

    Yasuaki Ikeda, Yuki Akura, Masaki Shimofuri, Amit Banerjee, Toshiyuki Tsuchiya, and Jun Hirotani. “Estimating Depth-Directional Thermal Conductivity Profiles Using Neural Network with Dropout in Frequency-Domain Thermoreflectance”. In: Journal of Applied Physics137.5 (Feb. 2025), p. 055106. (Visited on 05/09/2025)

    2025

  40. [40]

    Wide-Field Bond Quality Evaluation Using Frequency Domain Thermoreflectance with Deep Neural Network Feature Reconstruction

    Amun Jarzembski, Siddharth Nair, Wyatt Hodges, et al. “Wide-Field Bond Quality Evaluation Using Frequency Domain Thermoreflectance with Deep Neural Network Feature Reconstruction”. In:Advanced Materials Interfaces2401039.n/a (2025). (Visited on 05/15/2025)

    2025

  41. [41]

    Yuyao Ge, Lingrui Mei, Zenghao Duan, et al.A Survey of Vibe Coding with Large Language Models. Oct. 2025. (Visited on 06/25/2026)

    2025

  42. [42]

    May 2025

    Partha Pratim Ray.A Review on Vibe Coding: Fundamentals, State-of-the-art, Challenges and Future Directions. May 2025. (Visited on 06/25/2026)

    2025

  43. [43]

    Advait Sarkar and Ian Drosos.Vibe Coding: Programming through Conversation with Artificial Intelligence.Oct.2025.arXiv:2506.23253 [cs].(Visitedon06/25/2026)

    arXiv 2025

  44. [44]

    Uncertainty Analysis of Ther- moreflectance Measurements

    Jia Yang, Elbara Ziade, and Aaron J. Schmidt. “Uncertainty Analysis of Ther- moreflectance Measurements”. In:Review of Scientific Instruments87.1 (Jan. 2016), p. 014901.doi:10.1063/1.4939671

    doi:10.1063/1.4939671 2016

  45. [45]

    Version 0.1.x

    SST.OpenCode. Version 0.1.x. Open-source AI coding agent for the terminal. Ac- cessed: 2026-07-08. 2025.url:https://github.com/anomalyco/opencode. 29

    2026